Instruction Set Design

Transcription

Instruction Set Design
Instruction Set Design
• One goal of instruction set design is to minimize
instruction length
• Many instructions were designed with compilers in mind.
• Determining how operands are addressed is a key
component of instruction set design
Instruction Format
• Defines the layout of bits in an instruction
• Includes opcode and includes implicit or explicit
operand(s)
• Usually there are several instruction formats in an
instruction set
• Huge variety of instruction formats have been designed;
they vary widely from processor to processor
Instruction Length
• The most basic issue
• Affected by and affects:
– Memory size
– Memory organization
– Bus structure
– CPU complexity
– CPU speed
• Trade off between a powerful instruction repertoire and
saving space with shorter instructions
Instruction format trade-offs
Large instruction set => small programs
Small instruction set => large programs
Large memory => longer instructions
Fixed length instructions same size or multiple of bus
width => fast fetch
• Variable length instructions may need extra bus cycles
• Processor may execute faster than fetch
– Use cache memory or use shorter instructions
• Note complex relationship between word size, character
size, instruction size and bus transfer width
– In almost all modern computers these are all multiples
of 8 and related to each other by powers of 2
•
•
•
•
Allocation of bits
Determines several important factors
• Number of addressing modes
– Implicit operands don’t need bits
– X86 uses 2-bit mode field to specify Interpretation of 3bit operand fields
• Number of operands
– 3 operand formats are rare
– For two operand instructions we can use one or two
operand mode indicators
– X86 uses only one 2-bit indicator
Allocation of bits
Determines several important factors
• Register versus memory
– Tradeoff between # of registers and program size
– Studies suggest optimal number between 8 and 32
– Most newer architectures have 32 or more
– X86 architecture allows some computation in memory
• Number of register sets
– RISC architectures tend to have larger sets of uniform
registers
– Small register sets require fewer opcode bits
– Specialized register sets can reduce opcode bits further
by implicit reference (address vs. data registers)
Allocation of bits
Determines several important factors (cont’d)
• Address range
– Large address space requires large instructions for
direct addressing
– Many architectures have some restricted or short forms
of displacement addressing
Ex : x86 short jumps and loops,
PowerPC 16-bit displacement addressing
• Address granularity
– Size of object addressed
– Typically 8,16, 32 and 64 instruction variants
Addressing Modes
• addressing mode – method of forming a memory
address
• For a given instruction set architecture, addressing
modes define how machine language instructions
identify the operand (or operands) of each instruction.
• An addressing mode specifies how to calculate the
effective memory address of an operand by using
information held in registers and/or constants contained
within a machine instruction or elsewhere.
• Different types of addresses involve tradeoffs between
instruction length, addressing flexibility, and complexity
of address calculation
Addressing Modes
Common addressing modes
– Immediate
– Direct
– Indirect
– Register
– Register indirect
– Displacement
– Implied (stack)
Immediate Addressing
• the instruction itself contains the value to be used; located
in the address field of the instruction
• the value is stored in memory immediately after the
instruction opcode in memory
• Similar to using a constant in a high level language
• Advantage
– fast, since the value is included in the instruction; no
memory reference to fetch data
• Disadvantage
– not flexible, since the value is fixed at compile-time
– can have limited range in machines with fixed length
instructions
Instruction
operand
Immediate Addressing
for the following example, assume an accumulator machine structure and
that an add_immediate instruction is stored in memory, beginning at
location 12
memory
assembly lang
addr
contents
hardware actions
------------------------ ---------------------------------...
...
add_immediate(1)
12
| 41 |
acc
acc + 1
13
|1|
...
...
no additional memory
fetch for data beyond the
instruction fetch (since
the instruction contains
the data being used)
since an add must have different hardware actions than an add_immediate,
add_immediate has to be a different opcode (or there has to be an extra
type-of-addressing-mode code in the instruction format to go along with the
opcode)
Direct Addressing
• The instruction tells where the value can be found, but
the value itself is out in memory.
• The address field contains the address of the operand
• Effective address (EA) = address field (A)
• In a high level language, direct addressing is frequently
used for things like global variables.
• Advantage
– Single memory reference to access data
– More flexible than immediate
Instruction
MA
Memory
operand
Direct Addressing
for the following example, assume an accumulator machine structure and that
an add instruction is stored in memory, beginning at location 12
memory
assembly lang
addr
contents
hardware actions
--------------------------------------------------------...
...
add(one)
12
| 40 |
acc
acc + memory[24]
13
| 24 |
= acc + 1
...
...
word(one,1)
24
|1|
effective address = 24
...
...
so, when the PC points to 12:
40 (i.e., the contents of location 12) is interpreted as an opcode
24 (i.e., the contents of location 13) is interpreted as an address
1 (i.e., the contents of location 24) is interpreted as data
note that there are no tags or other indicators that the number 40 in location
12 has to be an opcode; it could just as well be used as an address or as data
Example of Immediate and Indirect Addressing Modes
Suppose we have a statement in C like
b = a + 10;
a and b are variables, so they are out in memory.
To execute this statement, we will need to fetch a from
memory, and write our result to b.
That means the instructions we generate need to have the
addresses of a and b, and need to read and write those
addresses as appropriate.
The number 10 is an actual value appearing in the statement.
So, our code needs to include 10 itself.
Memory-Indirect Addressing
• The memory cell pointed to by the address field contains
the address of (pointer to) the operand
• EA = (A)
Instruction
A
Memory
operand
Indirect Addressing
for the following examples, assume an accumulator machine
structure and that an add instruction is stored in memory,
beginning at location 12
assembly lang
------------------...
add_indirect(ptr)
...
word(one,1)
...
word(ptr,one)
...
memory
addr contents
hardware actions
------ --------------------------------...
12
| 42 | acc
acc + memory[memory[36]]
13
| 36 |
= acc + memory[24]
...
24
|1|
= acc + 1
...
36
| 24 | effective address = 24
...
the address included in the instruction is that of a pointer, that is, a
word that holds another address
Register Addressing
On machines with multiple registers, addresses and index values can be held in registers,
for example:
direct
load(x,r1)
// r1 <- memory[ x ]
immediate
load_imm(3,r2)
// r2 <- 3
indexed for array access
(fixed array base address
and index in a register)
load_ind(a,r3,r4)
// r4 <- memory[ a + r3 ]
register indirect as part
of indexed (i.e., a pointer
is in a register)
load_ind(0,r5,r6)
// r6 <- memory[ 0 + r5]
base plus displacemennt as
of indexed (i.e.,
structure access w/ ptr. in
reg. and constant offset)
load_ind(2,r7,r8)
// r8 <- memory[ 2 + r7 ] part
// accesses 3rd word of
// a structure
Register Addressing
• Operand(s) is (are) registers
• EA = R
– Register R is EA (not contents of R)
Instruction
R
operand
Registers
Register Addressing
• There is a limited number of registers
– A very small address field is needed
– Shorter instructions
– Faster instruction fetch
– X86: 3 bits used to specify one of 8 registers
• No memory access needed to fetch EA
• Very fast execution
• Very limited address space
• Multiple registers can help performance
• Requires good assembly programming or compiler writing
Note: in C you can specify register variables
register int a;
– This is only advisory to the compiler; no guarantees
Register-Indirect Addressing
• Similar to memory-indirect addressing
• EA = (R)
• Operand is in memory cell pointed to by contents of
register R
• Large address space (2n)
• One fewer memory address than memory-indirect
Instruction
R
Memory
operand
Registers
Displacement Addressing
• Combines register-indirect addressing and direct
addressing
• EA = A + (R)
• Address field holds two values
– A = base value
– R = register that holds displacement
– Or visa versa
Instruction
R
A
Memory
≈
operand
Registers
Types of Displacement Addressing
• Relative Addressing
• Base-register addressing
• Indexing
Relative Addressing
• EA = A + (PC)
• Address field A is treated as 2’s complement integer to allow
backward references
• Fetch operand from PC+A
• Can be very efficient because of locality of reference & cache usage
– But in large programs code and data may be widely separated in
memory
Base-Register Addressing
• A holds displacement
• R holds pointer to base address
• R may be explicit or implicit
– E.g. segment registers in 80x86 are base registers
and are involved in all EA computations
– X86 processors have a wide variety of base
addressing
Indexed Addressing
• A = Base
• R = displacement
• EA = A + R
• Good for accessing arrays
– EA = A + R
– R++
• Iterative access to sequential memory locations is very
common
• Some architectures provide auto-increment or autodecrement
• Preindex EA = A + (R++)
• Postindex EA = A + (++R)
• The ARM architecture provides preindexed and
postindexed addressing
Indexed Addressing
• Preindexed addressing
– Memory address is formed the same way as
displacement addressing, but the memory address is
written back to the base register after adding or
subtracting the displacement
– The writeback occurs before the store to memory
• Postindexed addressing
– Similar to preindexed addressing, but the writeback of
the effective address occurs after the store to
memory
Indexed Addressing Operands
• In addition to immediate values, the offset or
displacement can be in another register
• If a register is used then the addresses can be scaled
• The value in the offset register is scaled by one of the
shift operators
– Logical Shift Left / Right
– Arithmetic Shift Right
– Rotate Right
– Rotate Right Extended
• Amount of shift is an immediate operand in the
instruction
Indexed Addressing
for the following example, assume an accumulator machine structure and
that an add_indexed instruction is stored in memory, beginning at location 12
memory
assembly lang
addr contents
hardware actions
------------------------- ----------------------------------...
...
add_indexed(b0,x) 12
| 43 |
acc <- acc + memory[20+memory[36]]
13
| 20 |
= acc + memory[20+4]
14
| 36 |
= acc + memory[24]
...
...
= acc + 1
word(b0,5)
20 | 5 |
word(b1,-2)
21 | -2 |
effective address = 24
word(b2,3)
22 | 3 |
word(b3,9)
23 | 9 |
word(b4,1)
24 | 1 |
...
...
word(x,4)
36 | 4 |
...
...
Branch addressing modes
direct addressing, such as the accumulator machine
Assembly lang
addr
contents
hardware actions
--------------------------------------------------...
...
ba(target)
20
| 70 |
pc
30
21
| 30 |
...
...
label(target)
30
|
|
next instruction after branch
Branch addressing modes
pc-relative addressing, such as JVM
Assembly lang
-----------------...
goto(target)
...
label(target)
...
memory
addr
contents hardware actions
----------------- ---------------------...
20
| 167 |
pc
21 + 9
21
| 9 |
= 30
...
30
|
| next instruction after goto
...
note that other machines may make the offset relative to
the address of the branch (e.g., 20 above) or the fullyupdated pc (e.g., 22 above)
Stack Addressing
• Operand is implicitly on top of stack
– PUSH
– POP